ABSTRACT
This chapter first introduces advanced approaches for price segmentation based on latent regression models and multilevel regression models. It also introduces Gaussian mixture models as a variation on a theme of latent variable modeling. The type of dependent variable defines the type of regression model: ordinary least squared for a continuous dependent variable, logit for a discrete dependent variable, and Poisson for a count dependent variable. The chapter then describes multilevel models in terms of when they can be used, their complexities, and their relationship with the dummy variable model. It also utilizes the learning management system pricing scenario. JMP and Latent Class Gold are commercial software packages for estimating latent variable models. JMP only handles latent class models, while Latent Class Gold handles a wide array of models. The latent regression estimation was done using Latent Class Gold 4.5.
